CN110063694A - A kind of binocular sweeping robot and working method - Google Patents
A kind of binocular sweeping robot and working method Download PDFInfo
- Publication number
- CN110063694A CN110063694A CN201910347523.6A CN201910347523A CN110063694A CN 110063694 A CN110063694 A CN 110063694A CN 201910347523 A CN201910347523 A CN 201910347523A CN 110063694 A CN110063694 A CN 110063694A
- Authority
- CN
- China
- Prior art keywords
- sweeping robot
- binocular
- neural network
- processing
- mould group
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010408 sweeping Methods 0.000 title claims abstract description 112
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000012545 processing Methods 0.000 claims abstract description 77
- 238000013528 artificial neural network Methods 0.000 claims abstract description 50
- 238000004140 cleaning Methods 0.000 claims abstract description 43
- 230000000007 visual effect Effects 0.000 claims abstract description 17
- 238000013135 deep learning Methods 0.000 claims abstract description 16
- NJPPVKZQTLUDBO-UHFFFAOYSA-N novaluron Chemical compound C1=C(Cl)C(OC(F)(F)C(OC(F)(F)F)F)=CC=C1NC(=O)NC(=O)C1=C(F)C=CC=C1F NJPPVKZQTLUDBO-UHFFFAOYSA-N 0.000 claims description 13
- 230000004888 barrier function Effects 0.000 claims description 11
- 239000000428 dust Substances 0.000 claims description 8
- 230000000875 corresponding effect Effects 0.000 claims description 7
- 238000004458 analytical method Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000007726 management method Methods 0.000 claims description 5
- 238000010835 comparative analysis Methods 0.000 claims description 4
- 238000012546 transfer Methods 0.000 claims description 4
- 230000001133 acceleration Effects 0.000 claims description 3
- 238000009825 accumulation Methods 0.000 claims description 3
- 238000007373 indentation Methods 0.000 claims description 3
- 230000002093 peripheral effect Effects 0.000 claims description 3
- 239000007787 solid Substances 0.000 claims description 2
- 230000037237 body shape Effects 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 10
- 230000008859 change Effects 0.000 abstract description 3
- 238000013461 design Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 6
- 238000005259 measurement Methods 0.000 description 3
- 230000001276 controlling effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 210000000887 face Anatomy 0.000 description 2
- YREOLPGEVLLKMB-UHFFFAOYSA-N 3-methylpyridin-1-ium-2-amine bromide hydrate Chemical compound O.[Br-].Cc1ccc[nH+]c1N YREOLPGEVLLKMB-UHFFFAOYSA-N 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000010926 purge Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000007789 sealing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000930 thermomechanical effect Effects 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47L—DOMESTIC WASHING OR CLEANING; SUCTION CLEANERS IN GENERAL
- A47L11/00—Machines for cleaning floors, carpets, furniture, walls, or wall coverings
- A47L11/40—Parts or details of machines not provided for in groups A47L11/02 - A47L11/38, or not restricted to one of these groups, e.g. handles, arrangements of switches, skirts, buffers, levers
- A47L11/4011—Regulation of the cleaning machine by electric means; Control systems and remote control systems therefor
Abstract
The present invention relates to Smart Home technical fields, more particularly to a kind of binocular sweeping robot and working method, the binocular vision mould group of the binocular sweeping robot captures the figure of at least two angles to same object in real time, it carries out visual scanning and forms three-dimensional model, and positioning is carried out to cleaning region by control unit and builds figure and intelligent planning cleaning path.Using the working method of the binocular sweeping robot, the Processing with Neural Network module receives binocular vision mould group feedback information collected and carries out deep learning, so that sweeping robot only passes through binocular vision module, realize that visual scanning, slam positioning build the technical effect of figure and operating path planning, change the single visual theory application of tradition, the interior spatial structure of the highly integrated setting sweeping robot, design structure is reasonable, technical effect is prominent, is very suitable for the cleaning scene of sweeping robot.
Description
Technical field
The present invention relates to Smart Home technical fields, and in particular to a kind of binocular sweeping robot and working method.
Background technique
Currently, the family using sweeping robot is more and more.Sweeping robot currently on the market enter one it is new
It when environmental work, first has to traverse entire room according to algorithm, figure and positioning are built in completion.Then semantic map is constructed,
Path planning is finally carried out, then starts the cleaning to room again.Wherein, the method for traversing room is often sweeping robot one
It is directly walked close to a wall, after forming a closed circuit, then gradually fills up intermediate blank position, this way distance is longer,
It can devote a tremendous amount of time.
Equally, path planning algorithm is not only complicated, needs a large amount of calculating and human engineering, but also incomplete, cannot
Sweeping robot is set to work with optimal path.And after the camera of sweeping robot use on the market carries out Image Acquisition, only
Using image acquisition data carry out individual event visual scanning or slam positioning build figure and path planning, and identify avoidance still use compared with
For traditional physical sensing mode;
In recent years, depth enhancing study is quickly grown, and intelligent sweeping robot increased popularity, the past, robot was only able to achieve
Some simple tasks, but as artificial intelligence, the technology of neural network develop, it can preferably acquire and handle picture number
According to, and there are no such intelligent sweeping robot occur on the market at present.
Summary of the invention
In order to effectively solve the above problems, the present invention provides a kind of binocular sweeping robot and working method.
The specific technical solution of the present invention is as follows: a kind of binocular sweeping robot, and the binocular sweeping robot includes extremely
The ontology of few progress action cleans, at least one control unit fed back with Processing with Neural Network, at least one binocular
Vision mould group;
Described control unit is set in the ontology, and described control unit is connect with binocular vision mould group, the binocular
Vision mould group carries out visual scanning and forms three-dimensional model to the image of same at least two angle of object capture, and passes through control
Unit carries out positioning to cleaning region and builds figure and intelligent planning cleaning path.
Further, the binocular vision mould group includes at least one first camera and at least one second camera;
Described control unit includes that at least one controls processing module and at least one Processing with Neural Network module, described
First camera, second camera are all connect with Processing with Neural Network module, and the Processing with Neural Network module handles analysis
Result afterwards feeds back to the control processing module.
Further, at least one walking unit, at least one pressure sensitive unit, battery unit and at least one is clear
Clean unit;
Described control unit is connect with walking unit, pressure sensitive unit, battery unit and cleaning unit respectively.
Further, pressure sensitive unit is provided with below the binocular vision mould group, in preceding semicircle peripheral sides;
The pressure sensitive unit includes but is not limited to inductance pressure transducer, capacitance pressure transducer, potentiometer
Formula pressure sensor, Hall formula pressure sensor.
Further, the ontology includes at least one pedestal and at least one outer housing, the pedestal and outer housing phase
Mutually combination constitutes an oblate shape structural body;
The body interior is provided with the dirt box that can accommodate rubbish, dust, at least one is arranged on the top surface of the ontology
Dirt box cover.
Further, described control unit further includes at least one control button, and the control button is arranged at described
On the top surface of body, and the control button and control processing module are connected with each other;
The control processing module includes but is not limited to one-chip computer module;
Processing with Neural Network module includes but is not limited to that can accumulate the embedded nerve of identification type of goods with deep learning
Network (NPU) processor, Intel's neural network processor.
Further, the walking unit includes the first travel wheel, the second travel wheel, universal wheel, electric machine assembly, and described the
One travel wheel, the second travel wheel are symmetrically arranged on the bottom surface of pedestal, first travel wheel, the second travel wheel all with electricity
Thermomechanical components are connected with each other;
The bottom surface of the pedestal is provided at least one universal wheel at the position of pressure sensitive unit;
The battery unit includes that at least one can the batteries of charge and discharge, at least one charging that can be charged the battery
Seat, the battery are connect with control unit, and the cradle is fixed on some sweeping robot accessible position.
A kind of working method of binocular sweeping robot, the working method the following steps are included:
A1 starts work: binocular vision mould group work, cooperates rotation in sweeping robot traveling, shoots overall space, answer
Visual scanning is carried out to same object with the first camera, second camera, obtains the first image, the second image in real time, is constituted
Three-dimensional model;
Figure is built in A2 positioning: right by the picture transfer of binocular vision mould group shooting to the Processing with Neural Network module
Real time picture disparity computation directly obtains each point distance of forward image, establishes the point cloud chart of space environment, utilize point cloud chart
Come while picking up, projection constitutes the map in a complete separate room again;
A3 path planning: sweeping robot control unit builds up target map completely, calculates ground by predefined paths algorithm
Target map is divided into multiple regions, carries out cleaning by the optimal case of figure cleaning inside;
A4 recharges continuation of the journey: sweeping robot is set up by binocular vision mould group captured in real-time photo, and to the subject bodily form
Shape, constantly the cradle picture with the storage of Processing with Neural Network module carry out model comparison, obtain the position of cradle, and
Return charging;
A5 resets work: after the completion of sweeping robot cleans, executing default program and is to revert on cradle, entrance is standby
Dormant state waits the preset starting of working time next time or manually starts.
Further, figure is built in the A2 positioning further include:
B1 constructs preliminary map: if sweeping robot moves to the place of door or notch, B2 is thened follow the steps, if
Complete preliminary cartographic information is constructed in Processing with Neural Network module, thens follow the steps B3;
B2 continues positioning and builds figure: in the place for having door or notch, sweeping robot enters another from door or indentation, there
A unit continues shooting and establishes space map, repeats the above steps, and until completing an enclosure space map, composition one is complete
The preliminary map in whole, closed cleaning region;
B3 adjustment map: after preliminary map establishes storage in memory, sweeping robot starts to walk along wall, machine
The angular speed and acceleration of interior IMU measurement ontology, the practical walking route of recorder record the foot of a wall or furniture when cleaning
Position, and pass through the deep learning of Processing with Neural Network module, it compares, establishes complete with the preliminary map of storage in real time
Target map.
Further, the A3 path planning further include:
C1 emergent management: sweeping robot keeps off barrier in front, passes through sweeping robot first half for occurring suddenly
The pressure sensitive unit of circle periphery forms feedback signal, is transferred to control unit after hitting the slight deformation for squeezing and generating,
The signal has highest priority, and described control unit provides electric machine assembly reverse signal, after sweeping robot can be realized
Keep out of the way barrier, executes step S7;
C2 is relearned: sweeping robot adjusts the angle, and the binocular vision mould group shoots obstructions chart picture again, carries out
Ranging calculates, and the prototype comparative analysis with the storage of Processing with Neural Network module is handled, and processing result is transferred to control processing
Module, the control processing module is according to feedback, then does corresponding corresponding actions processing, and the Processing with Neural Network module is urgent
Deep learning in treatment process, the type of accumulation identification article;
C3 is especially avoided: during sweeping robot cleans ground, being encountered small articles, is known by Processing with Neural Network module
After not, preferential evacuation detours.
Usefulness of the present invention: a kind of binocular sweeping robot of the present invention and working method, machine of sweeping the floor are applied
The image information that people obtains according to binocular vision mould group, and principle of parallax is applied, three-dimensional model is constructed to each image, is passed through
The deep learning of neural network can judge in time the variation of ground environment to be cleaned, and control the Working mould of sweeping robot
Formula follows the ground environment of cleaning to be changed, so that sweeping robot faces different ground environments, can guarantee cleaning effect
Fruit;
Further, the Processing with Neural Network module receiving binocular vision mould group feedback information collected is gone forward side by side
Row deep learning realizes that visual scanning, slam positioning build figure and work so that sweeping robot only passes through binocular vision module
The technical effect of path planning changes the single visual theory application of tradition, highly integrated that the interior of the sweeping robot is arranged
Portion's space structure, design structure is reasonable, and technical effect is prominent, is very suitable for the cleaning scene of sweeping robot.
Detailed description of the invention
Fig. 1 is the overall structure diagram of first embodiment of the invention;
Fig. 2 is that the overall structure of first embodiment of the invention splits schematic diagram;
Fig. 3 is the connection relationship diagram of first embodiment of the invention;
Fig. 4 is the overall structure diagram at another visual angle of first embodiment of the invention;
Fig. 5 is the workflow schematic diagram of second embodiment of the invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is explained in further detail.It should be appreciated that specific embodiment described herein is used only for explaining the present invention, and
It is not used in the restriction present invention.
On the contrary, the present invention covers any substitution done on the essence and scope of the present invention being defined by the claims, repairs
Change, equivalent method and scheme.Further, in order to make the public have a better understanding the present invention, below to of the invention thin
It is detailed to describe some specific detail sections in section description.Part without these details for a person skilled in the art
The present invention can also be understood completely in description.
It as shown in Figure 1 and Figure 2, is the overall structure diagram of first embodiment of the invention, this embodiment offers a kind of double
Mesh sweeping robot, the binocular sweeping robot include at least one ontology 1, at least one control unit 3, at least one pair
Visually feel mould group 4, at least one walking unit 5, at least one pressure sensitive unit 6, at least one cleaning unit 7 and at least
One battery unit;
In the present embodiment, the ontology 1 of the binocular sweeping robot uses oblate shape structural body, and the ontology 1 wraps
Include at least one pedestal 10 and at least one outer housing 11, the pedestal 10 and outer housing 11 can be combined with each other into one it is oblate
Shape structural body, the combination settings mode both include but is not limited to mutually be spirally connected, be clamped, the those skilled in the art such as snapping it is common
Fixed form.
It is internally provided with the dirt box that can accommodate rubbish, dust in the ontology 1, does not illustrate dirt box structure in figure, it is described
Dirt box is the conventional equipment of the sweeping robot of this field, and position and setting relationship of the dirt box in ontology 1 are this field
Conventional equipment, be not specifically limited herein;
At least one dirt box cover 2 is set on the top surface of the ontology 1, and the dirt box cover 2 is for dirt described in temporary sealing
Box guarantees that rubbish, dust are temporarily stored in dirt box, and the dirt box cover 2 is arranged in the top side location of ontology 1, convenient for using
It directly opens dirt box cover 2 and carries out clean rubbish in family.
Described control unit 3 is set on the top surface of the ontology 1, and in the present embodiment, described control unit 3 includes extremely
A few control button, at least one control processing module and at least one Processing with Neural Network module, the control button are set
It sets on 1 top surface of ontology, in the present embodiment, the control button includes but is not limited to use push-button switch;
As shown in figure 3, the control processing module is arranged inside ontology 1, the control processing module is for controlling simultaneously
It handles the data of the sweeping robot feedback and controls the motion mode of sweeping robot, the control processing module and control
System realizes the operating interactive with user, the control processing module and walking unit by key connection, and by the control button
5, pressure sensitive unit 6, cleaning unit 7, battery unit are separately connected, to control modules collaborative work, complete to sweep the floor
The intelligent cleaning of robot;The control processing module includes but is not limited to one-chip computer module, does not do specific limit herein
It is fixed;
Further, described control unit 3 is integrally disposed the Processing with Neural Network module for deep learning, institute
It states Processing with Neural Network module and control processing module is connected with each other, carry out information exchange communication, the Processing with Neural Network mould
Block receives the binocular vision mould group 4 feedback information collected and simultaneously carries out deep learning so that sweeping robot only pass through it is double
Mesh vision module realizes that visual scanning, slam positioning build the technical effect of figure and operating path planning.
In the present embodiment, Processing with Neural Network module includes but is not limited to that can accumulate identification article kind with deep learning
Embedded neural network (NPU) processor of class, Intel's neural network processor etc., are not specifically limited herein;
Specifically, binocular vision mould group 4 is provided on the side of the direction of advance of the ontology 1, in the present embodiment,
The binocular vision mould group 4 includes the first camera 41, second camera 42, first camera 41, second camera 42
It is all connect with Processing with Neural Network module, and acquired image information input Processing with Neural Network module is subjected to depth
It practises and the feedback arrangement after processing analysis is inputted the control processing module by processing analysis, the Processing with Neural Network module
In, as the movement foundation of control processing module, and according to built-in operating path, controls the walking unit 5 and moved;
Image Acquisition is carried out to an object simultaneously by first camera 41, second camera 42, is carried out to two
The calculating of width image parallactic, it is understood that it is that range measurement directly is carried out to front scenery (range taken by image),
Without judging that front occurs that kind of barrier.So for any kind of barrier, it can be according to distance
The variation of information carries out necessary early warning or braking.
Since the principle of binocular camera is similar to human eye, human eye can perceive the distance of object, be due to two eyes
The image presented to the same object has differences, also referred to as " parallax ".Object distance is remoter, and parallax is smaller;Conversely, parallax is got over
Greatly.The size of parallax corresponds to the distance of distance between object and eyes, this is also that 3D film can make one solid layer feeling
The reason of knowing.
The binocular vision mould group 4 can obtain indoor 3 D stereo map when sweeping robot carries out visual scanning,
And according to the positional relationship of object, carries out slam positioning and build figure, when needing to carry out cleaning, equally pass through the binocular vision
Feel that mould group 4 carries out operating path planning, it is ensured that each position in cleaning region is cleaned, and the binocular vision is made full use of
Feel the principle of parallax of mould group 4, so that 1 inner space of ontology is highly integrated, process of sweeping the floor is more intelligent.
Further, it is provided with pressure sensitive unit 6 below the binocular vision mould group 4, in preceding semicircle peripheral sides,
The pressure sensitive unit 6 is used to detect the barrier under emergency case, and the pressure sensitive unit 6 includes in the present embodiment
But it is not limited to inductance pressure transducer, capacitance pressure transducer, potentiometer-type pressure transducer, Hall formula pressure sensor
Deng pressure sensor common to those skilled in the art is not specifically limited herein;
Occur in sweeping robot in face of suddenly, barrier of the gear before machine, the people such as to pass by, or specially close
Object, be in map it is unwritten, machine can not in time by the camera shooting analysis processing of binocular mould group when, pass through ontology 1
The pressure sensor of semicircle periphery before shell forms feedback signal, is transferred to control after hitting the slight deformation for squeezing and generating
Processing module;
Described control unit 3 provides running gear reverse signal, and the sweeping robot can retreat, then adjusting angle
Degree, binocular vision mould group 4 continue to shoot upload process center, the prototype after ranging calculates, with the storage of Processing with Neural Network module
Comparative analysis, then processing feedback is carried out to control processing module, thus scientific carry out respective handling again, or continue to retreat,
Or it detours, or move on, (barrier is withdrawn).
As shown in figure 4, in the present embodiment, the walking unit 5 includes the first travel wheel 51, the second travel wheel 52, ten thousand
To wheel 53, electric machine assembly 54, first travel wheel 51, the second travel wheel 52 are symmetrically arranged on the bottom surface of pedestal 10,
Corresponding first travel wheel 51 of the pedestal 10, the second travel wheel 52 are through offering lead to the hole site, first travel wheel 51, the
Two travel wheels 52 are all connected with each other with electric machine assembly 54, and are that this field can drive described the with the electric machine assembly 54 is cleaned
The conventional motor device that one travel wheel 51, the second travel wheel 52 are rotated, is not specifically limited herein;
The middle position of first travel wheel 51, the second travel wheel 52, it is understood that be the bottom surface of the pedestal 10
It is provided at least one universal wheel 53 at the position of pressure sensitive unit 6, the universal wheel 53 is by another small electrical unit
Part 54 drives, and the universal wheel 53 can carry out divertical motion under the driving of the motor, the motion mode of the universal wheel 53,
And driving principle uses the transfer of this field routine, the entire walking unit 5 only provides the sweeping robot
The movement of cleaning track is realized, not the core technology scheme of the technical problem of being solved of the technical program, in other implementations
In example, only to realize that the sweeping robot can meet the normal requirement for realizing the movement of cleaning track, does not do have herein
Body limits;
The cleaning unit 7 includes at least one dust suction subassembly, and the dust suction subassembly is by sweeping robot by cleaning rail
The rubbish of mark focuses on, and temporarily stores into dirt box, and the dust suction subassembly is the conventional dust exhaust apparatus of this field, only with reality
The requirement on existing sweeping robot cleaning ground, does not limit specifically herein;
The battery unit include at least one can charge and discharge battery, the battery is connect with control unit 3, and is provided
Each unit module work normally electric energy, the battery unit be this field routine can charge and discharge cell apparatus, herein
It is not specifically limited;
The battery unit further includes the cradle that at least one can be charged the battery, the cradle fixed setting
On some sweeping robot accessible position, in sweeping robot work, if control unit 3 detects battery capacity not
When sufficient, sweeping robot begins look for cradle, shoots photo by binocular vision mould group 4, forms stereoscopic three-dimensional shape in real time,
And it is constantly compared with the interior charging Cuo picture set.When the comparison registration of two models reaches predetermined value, the machine of sweeping the floor
People's adjustment direction is advanced, and after charging pole connects, stop motion starts to charge the battery;
Further, after the completion of to be charged, sweeping robot returns the unfinished point of cleaning, continues to complete by planning clear
It sweeps;After the completion of cleaning, default program is to revert on cradle, charging, after being full of, into standby dormant state.It waits lower default
Working time next time starting or manually start.
As shown in figure 5, the second embodiment provides a kind of binocular sweeping robot in the second embodiment of the present invention
Working method, the working method using in above-mentioned sweeping robot, and the working method specifically includes the following steps:
S1 starts work: opening sweeping robot power switch, binocular vision mould group 4 works, in sweeping robot traveling
Cooperation rotation, shoots overall space, carries out visual scanning to same object using the first camera 41, second camera 42, real
When obtain the first image, the second image, constitute three-dimensional model;
Figure is built in S2 positioning: the picture transfer that the binocular vision mould group 4 is shot is right to the Processing with Neural Network module
Real time picture disparity computation directly obtains each point distance of forward image, establishes the point cloud chart of space environment, utilize point cloud chart
Come while picking up, projection constitutes the map in a complete separate room again, if sweeping robot has moved to door or notch, (width is big
In machine itself) place, S3 is thened follow the steps, if constructing complete preliminary map letter in Processing with Neural Network module
Breath, thens follow the steps S4;
S3 continues positioning and builds figure: in the place for having door or notch (width is greater than machine itself), sweeping robot is from door
Or indentation, there enters another unit, continues shooting and establishes space map, repeats the above steps, it is empty until completing a closing
Between map, constitute one it is complete, it is closed cleaning region preliminary map;
S4 adjustment map: after preliminary map establishes storage in memory, sweeping robot starts to walk along wall, machine
The angular speed and acceleration of interior IMU (inertial sensor) measurement ontology 1, the practical walking route of recorder are remembered when cleaning
The foot of a wall (or the profiles such as furniture) position is recorded, and passes through the deep learning of Processing with Neural Network module, in real time preliminarily with storage
Figure compares, and establishes complete target map;
S5 intelligent planning cleaning path: sweeping robot along wall cleaning after the completion of, sweeping robot control unit 3
Map is built up completely, and the optimal case of map cleaning inside is calculated by predefined paths algorithm, target map is divided into multiple
Region only reduces machine evacuation to realize as far as possible, turns round, turn around and be overlapped the technical effect of purging zone;
Or user can including but not limited to have mobile phone, the tablet computer etc. of visual interface by intelligent terminal, into
The customized manual allocation of row cleans region, to meet customized requirement for cleaning;
S6 emergent management: sweeping robot keeps off barrier in front, passes through sweeping robot first half for occurring suddenly
The pressure sensitive unit 6 of circle periphery forms feedback signal, is transferred to control unit after hitting the slight deformation for squeezing and generating
3, which has highest priority, and described control unit 3 provides 54 reverse signal of electric machine assembly, and machine of sweeping the floor can be realized
People retreats avoidance, executes step S7;
S7 is relearned: sweeping robot adjusts the angle, and the binocular vision mould group 4 shoots obstructions chart picture again, into
Row ranging calculates, and the prototype comparative analysis with the storage of Processing with Neural Network module is handled, and processing result is transferred at control
Module is managed, the control processing module is according to feedback, then does corresponding corresponding actions processing, and the Processing with Neural Network module is tight
Deep learning in anxious treatment process, the type of accumulation identification article;
In the present embodiment, during sweeping robot cleans, the barrier encountered includes but is not limited to the people to pass by,
Or specially close object, be in map it is unwritten, sweeping robot can not imaged by binocular mould group in time
When analysis processing, emergent management step can be started;
S8 is especially avoided: during sweeping robot cleans ground, encountering small articles, (volume ratio machine is small, enters than dirt box
Mouthful big, be highly no more than machine itself) identified by Processing with Neural Network module after, preferential evacuation detours.
The preset cradle of S9: the sweeping robot further includes at least one cradle, and the cradle is arranged in vision
In the body of a map or chart of scanning, in sweeping robot work, if thening follow the steps S10 when electric power detection deficiency;
S10 recharges continuation of the journey: sweeping robot begins look for cradle, by 4 captured in real-time photo of binocular vision mould group, and
Three-dimensional shape is formed to subject body, the constantly cradle picture with the storage of Processing with Neural Network module carries out model comparison;
If comparison registration reaches, sweeping robot adjustment direction advances and after charging pole connects, stop motion, into
Row charging continuation of the journey after the completion of to be charged, returns the unfinished point of cleaning, continues to complete to clean by planning;
S11 resets work: after the completion of sweeping robot cleans, executing default program and is to revert on cradle, charge, fill
Man Hou waits the preset starting of working time next time or manually starts into standby dormant state.
The figure obtained using the working method of the binocular sweeping robot, sweeping robot according to binocular vision mould group 4
As information, and principle of parallax is applied, three-dimensional model is constructed to each image, it, can be timely by the deep learning of neural network
The operating mode for judging the variation of ground environment to be cleaned, and controlling sweeping robot follows the ground environment of cleaning to be become
Change, so that sweeping robot faces different ground environments, can guarantee cleaning effect;
Further, the Processing with Neural Network module receives the binocular vision mould group 4 feedback information collected simultaneously
Deep learning is carried out, so that sweeping robot only passes through binocular vision module, realizes that visual scanning, slam positioning build figure and work
Make the technical effect of path planning, changes the single visual theory application of tradition, the highly integrated setting sweeping robot
Interior spatial structure, design structure is reasonable, and technical effect is prominent, is very suitable for the cleaning scene of sweeping robot.
For the ordinary skill in the art, introduction according to the present invention, do not depart from the principle of the present invention with
In the case where spirit, changes, modifications, replacement and the deformation that embodiment is carried out still fall within protection scope of the present invention it
It is interior.
Claims (10)
1. a kind of binocular sweeping robot, which is characterized in that the binocular sweeping robot includes that at least one move clearly
Clean ontology, at least one control unit, at least one binocular vision mould group with Processing with Neural Network feedback;
Described control unit is set in the ontology, and described control unit is connect with binocular vision mould group, the binocular vision
Mould group carries out visual scanning and forms three-dimensional model to the image of same at least two angle of object capture, and passes through control unit
Positioning is carried out to cleaning region and builds figure and intelligent planning cleaning path.
2. a kind of binocular sweeping robot and working method according to claim 1, which is characterized in that the binocular vision mould
Group includes at least one first camera and at least one second camera;
Described control unit includes at least one control processing module and at least one Processing with Neural Network module, and described first
Camera, second camera are all connect with Processing with Neural Network module, and treated by analysis for the Processing with Neural Network module
As a result the control processing module is fed back to.
3. a kind of binocular sweeping robot and working method according to claim 1, which is characterized in that at least one walking is single
Member, at least one pressure sensitive unit, battery unit and at least one cleaning unit;
Described control unit is connect with walking unit, pressure sensitive unit, battery unit and cleaning unit respectively.
4. a kind of binocular sweeping robot and working method according to claim 2, which is characterized in that in the binocular vision
Pressure sensitive unit is provided with below mould group, in preceding semicircle peripheral sides;
The pressure sensitive unit includes but is not limited to inductance pressure transducer, capacitance pressure transducer, potentiometer type pressure
Force snesor, Hall formula pressure sensor.
5. a kind of binocular sweeping robot and working method according to claim 1, which is characterized in that the ontology includes extremely
A few pedestal and at least one outer housing, the pedestal and outer housing, which are combined with each other, constitutes an oblate shape structural body;
The body interior is provided with the dirt box that can accommodate rubbish, dust, at least one dirt box is arranged on the top surface of the ontology
Lid.
6. a kind of binocular sweeping robot and working method according to claim 2, which is characterized in that described control unit is also
Including at least one control button, the control button is arranged on the top surface of the ontology, and the control button and control
Processing module is connected with each other;
The control processing module includes but is not limited to one-chip computer module;
Processing with Neural Network module includes but is not limited to that can accumulate the embedded neural network of identification type of goods with deep learning
(NPU) processor, Intel's neural network processor.
7. a kind of binocular sweeping robot and working method according to claim 3, which is characterized in that the walking unit packet
The first travel wheel, the second travel wheel, universal wheel, electric machine assembly are included, first travel wheel, the second travel wheel are symmetrically arranged
On the bottom surface of pedestal, first travel wheel, the second travel wheel are all connected with each other with electric machine assembly;
The bottom surface of the pedestal is provided at least one universal wheel at the position of pressure sensitive unit;
The battery unit include at least one can charge and discharge battery, at least one cradle that can be charged the battery,
The battery is connect with control unit, and the cradle is fixed on some sweeping robot accessible position.
8. a kind of working method of binocular sweeping robot, the working method the following steps are included:
A1 starts work: binocular vision mould group work, sweeping robot advance in cooperate rotation, overall space is shot, using the
One camera, second camera carry out visual scanning to same object, obtain the first image, the second image in real time, constitute three-dimensional
Model;
Figure is built in A2 positioning: by the picture transfer of binocular vision mould group shooting to the Processing with Neural Network module, to real-time
Picture disparity computation is directly obtained each point distance of forward image, establishes the point cloud chart of space environment, connect using point cloud chart side
Get up, projection constitutes the map in a complete separate room again;
A3 path planning: sweeping robot control unit builds up target map completely, is calculated in map by predefined paths algorithm
Target map is divided into multiple regions, carries out cleaning by the clean optimal case in portion;
A4 recharges continuation of the journey: sweeping robot forms solid figure by binocular vision mould group captured in real-time photo, and to subject body
Shape, constantly the cradle picture with the storage of Processing with Neural Network module carry out model comparison, obtain the position of cradle, and return
Charging;
A5 resets work: after the completion of sweeping robot cleans, executing default program and is to revert on cradle, into standby suspend mode
State waits the preset starting of working time next time or manually starts.
9. a kind of working method of binocular sweeping robot according to claim 8, which is characterized in that figure is built in the A2 positioning
Further include:
B1 constructs preliminary map: if sweeping robot moves to the place of door or notch, B2 is thened follow the steps, if in mind
Through constructing complete preliminary cartographic information in network process module, B3 is thened follow the steps;
B2 continues positioning and builds figure: in the place for having door or notch, sweeping robot enters another list from door or indentation, there
Member continues shooting and establishes space map, repeats the above steps, and until completing an enclosure space map, constitutes a complete, envelope
The preliminary map in the cleaning region closed;
B3 adjustment map: after preliminary map establishes storage in memory, sweeping robot starts to walk along wall, in machine
IMU measures the angular speed and acceleration of ontology, and the practical walking route of recorder records the foot of a wall or furniture position when cleaning,
And pass through the deep learning of Processing with Neural Network module, it is compared in real time with the preliminary map of storage, establishes complete target
Map.
10. a kind of working method of binocular sweeping robot according to claim 8, which is characterized in that the path the A3 rule
It draws further include:
C1 emergent management: sweeping robot keeps off barrier in front for occurring suddenly, by semicircle before sweeping robot outside
The pressure sensitive unit enclosed forms feedback signal, is transferred to control unit, the letter after hitting the slight deformation for squeezing and generating
Number there is highest priority, and described control unit provides electric machine assembly reverse signal, keeps out of the way after sweeping robot can be realized
Barrier executes step S7;
C2 is relearned: sweeping robot adjusts the angle, and the binocular vision mould group shoots obstructions chart picture again, carries out ranging
It calculating, the prototype comparative analysis with the storage of Processing with Neural Network module is handled, and processing result is transferred to control processing module,
The control processing module is according to feedback, then does corresponding corresponding actions processing, and the Processing with Neural Network module is in emergent management
Deep learning in the process, the type of accumulation identification article;
C3 is especially avoided: during sweeping robot cleans ground, being encountered small articles, is identified by Processing with Neural Network module
Afterwards, preferential evacuation detours.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910347523.6A CN110063694A (en) | 2019-04-28 | 2019-04-28 | A kind of binocular sweeping robot and working method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910347523.6A CN110063694A (en) | 2019-04-28 | 2019-04-28 | A kind of binocular sweeping robot and working method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110063694A true CN110063694A (en) | 2019-07-30 |
Family
ID=67369244
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910347523.6A Pending CN110063694A (en) | 2019-04-28 | 2019-04-28 | A kind of binocular sweeping robot and working method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110063694A (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110499727A (en) * | 2019-08-14 | 2019-11-26 | 北京智行者科技有限公司 | A kind of welt cleaning method and sweeper based on multisensor |
CN110825088A (en) * | 2019-11-29 | 2020-02-21 | 燕山大学 | Multi-view vision guiding ship body cleaning robot system and cleaning method |
CN112401752A (en) * | 2020-11-04 | 2021-02-26 | 北京石头世纪科技股份有限公司 | Method, device, medium and electronic equipment for detecting unknown obstacles |
CN112587036A (en) * | 2020-12-10 | 2021-04-02 | 苏州阿甘机器人有限公司 | Unmanned sweeper based on machine vision and working method thereof |
CN113534812A (en) * | 2021-07-30 | 2021-10-22 | 上海高仙自动化科技发展有限公司 | Cleaning robot control system and cleaning robot |
CN113729561A (en) * | 2021-09-26 | 2021-12-03 | 复旦大学 | Cleaning robot and control method thereof |
CN113885519A (en) * | 2021-10-27 | 2022-01-04 | 北京小乔机器人科技发展有限公司 | Method for controlling robot to automatically follow |
WO2022041737A1 (en) * | 2020-08-28 | 2022-03-03 | 北京石头世纪科技股份有限公司 | Distance measuring method and apparatus, robot, and storage medium |
CN114415657A (en) * | 2021-12-09 | 2022-04-29 | 安克创新科技股份有限公司 | Cleaning robot wall-following method based on deep reinforcement learning and cleaning robot |
CN114794953A (en) * | 2022-04-19 | 2022-07-29 | 净豹智能机器人(台州)有限公司 | Intelligent cleaning robot welt cleaning system and method based on deep learning |
CN114913223A (en) * | 2021-02-09 | 2022-08-16 | 北京盈迪曼德科技有限公司 | Positive direction identification method and system of visual sweeper |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2861855A1 (en) * | 2003-11-03 | 2005-05-06 | Wany Sa | METHOD AND DEVICE FOR SCANNING A SURFACE STATISTICALLY |
CN107092254A (en) * | 2017-04-27 | 2017-08-25 | 北京航空航天大学 | A kind of design method for the Household floor-sweeping machine device people for strengthening study based on depth |
CN108888187A (en) * | 2018-05-31 | 2018-11-27 | 四川斐讯信息技术有限公司 | A kind of sweeping robot based on depth camera |
-
2019
- 2019-04-28 CN CN201910347523.6A patent/CN110063694A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FR2861855A1 (en) * | 2003-11-03 | 2005-05-06 | Wany Sa | METHOD AND DEVICE FOR SCANNING A SURFACE STATISTICALLY |
CN107092254A (en) * | 2017-04-27 | 2017-08-25 | 北京航空航天大学 | A kind of design method for the Household floor-sweeping machine device people for strengthening study based on depth |
CN108888187A (en) * | 2018-05-31 | 2018-11-27 | 四川斐讯信息技术有限公司 | A kind of sweeping robot based on depth camera |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110499727A (en) * | 2019-08-14 | 2019-11-26 | 北京智行者科技有限公司 | A kind of welt cleaning method and sweeper based on multisensor |
CN110825088A (en) * | 2019-11-29 | 2020-02-21 | 燕山大学 | Multi-view vision guiding ship body cleaning robot system and cleaning method |
WO2022041737A1 (en) * | 2020-08-28 | 2022-03-03 | 北京石头世纪科技股份有限公司 | Distance measuring method and apparatus, robot, and storage medium |
CN112401752A (en) * | 2020-11-04 | 2021-02-26 | 北京石头世纪科技股份有限公司 | Method, device, medium and electronic equipment for detecting unknown obstacles |
WO2022095488A1 (en) * | 2020-11-04 | 2022-05-12 | 北京石头创新科技有限公司 | Method and apparatus for detecting unknown obstacle, and medium and electronic device |
CN112587036A (en) * | 2020-12-10 | 2021-04-02 | 苏州阿甘机器人有限公司 | Unmanned sweeper based on machine vision and working method thereof |
CN114913223A (en) * | 2021-02-09 | 2022-08-16 | 北京盈迪曼德科技有限公司 | Positive direction identification method and system of visual sweeper |
CN113534812A (en) * | 2021-07-30 | 2021-10-22 | 上海高仙自动化科技发展有限公司 | Cleaning robot control system and cleaning robot |
CN113729561A (en) * | 2021-09-26 | 2021-12-03 | 复旦大学 | Cleaning robot and control method thereof |
CN113885519A (en) * | 2021-10-27 | 2022-01-04 | 北京小乔机器人科技发展有限公司 | Method for controlling robot to automatically follow |
CN114415657A (en) * | 2021-12-09 | 2022-04-29 | 安克创新科技股份有限公司 | Cleaning robot wall-following method based on deep reinforcement learning and cleaning robot |
CN114794953A (en) * | 2022-04-19 | 2022-07-29 | 净豹智能机器人(台州)有限公司 | Intelligent cleaning robot welt cleaning system and method based on deep learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110063694A (en) | A kind of binocular sweeping robot and working method | |
CN205989678U (en) | A kind of unpiloted sweeper | |
CN111990929B (en) | Obstacle detection method and device, self-walking robot and storage medium | |
TWI654961B (en) | Robotics device, terminal device, system for remotely controlling robotics device, and program | |
KR101366860B1 (en) | Mobile robot and controlling method of the same | |
CN109431381A (en) | Localization method and device, electronic equipment, the storage medium of robot | |
EP3705010B1 (en) | Suction collecting station, system comprising a suction collecting station and a suction cleaner and method for same | |
EP3494446A1 (en) | Method for controlling an autonomous mobile robot | |
CN110063687A (en) | Self-propelled electric dust collector | |
JP2019528535A (en) | Mobile robot and control method thereof | |
US20190220033A1 (en) | Moving robot and controlling method for the moving robot | |
JP2003180587A (en) | Electric cleaner with detachable unit | |
CN110325938A (en) | Electric dust collector | |
CN110376934A (en) | Clean robot, clean robot control method and terminal control method | |
AU2014278987A1 (en) | Cleaning robot and method for controlling the same | |
CN105209997A (en) | Apparatus for guiding an autonomous vehicle towards a docking station | |
CN104541517A (en) | Device for creation of layout information, method for creation of layout information, system for operation of domestic electrical appliances, and self-propelled electronic device | |
CN211022482U (en) | Cleaning robot | |
CN106239517A (en) | Robot and the method for the autonomous manipulation of realization, device | |
CN108544912A (en) | Four-wheel differentia all-terrain mobile robot control system and its control method | |
CN109634267A (en) | One kind being used for shopping mall supermarket intelligent picking distributed robot | |
CN113378684B (en) | Cleaning device and light trigger event identification method for cleaning device | |
CN211786765U (en) | Intelligent indoor autonomous navigation robot | |
EP3555721A1 (en) | Method for creating an environment map for a processing unit | |
CN109938642A (en) | Electric dust collector |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |